iiinova — where your digital self keeps evolving

Prof. Benjamin Forest @benjaminforest.bsky.social · 02/09/2026

Nouvelle prévision basée sur un sondage Léger du 31 août (presque identique au sondage Pallas Data du 29). Le PQ est proche de la majorité et le PLQ semble avoir une nette avance sur la CAQ. Peu de changement pour le QS et le CPQ.

000

Prof. Benjamin Forest @benjaminforest.bsky.social · 30/08/2026

Merci à @joffre.bsky.social d'avoir repéré une erreur flagrante dans mes prévisions. (Dans mon code, j'avais inversé les chiffres du vote libéral de 2022 : 41,4 % au lieu de 14,4 %.) Cela a amené le modèle à sous-estimer fortement le nombre de sièges libéraux. Voici le graphique corrigé pour Leger.

010

Prof. Benjamin Forest @benjaminforest.bsky.social · 30/08/2026

Stop the presses! Thanks to @joffre.bsky.social for identifying an obvious error in my forecast. (In my code, I transposed the digits of 2022 Liberal vote as 41.4% rather than the correct 14.4%.) That led the model to severely under-predict Liberal seats. Here's the corrected graph for Synopsis.

110

Prof. Benjamin Forest @benjaminforest.bsky.social · 30/08/2026

Le sondage Léger du 24 août indiquait un appui plus important à la CAQ que le sondage Synopsis réalisé quatre jours plus tard. Ces résultats confèrent à la CAQ une confortable majorité.

000

Prof. Benjamin Forest @benjaminforest.bsky.social · 29/08/2026

The next Quebec election is October 5 so it's time to fire up the swing model again. This is based on the 28 August poll by Synopsis, which is more favourable to the CAQ than other recent surveys. Based on it, the CAQ and PQ are neck-and-neck but no party is likely to win a majority of seats.

1113

Prof. Benjamin Forest @benjaminforest.bsky.social · 18/05/2026

Georgia had four "Alito" counties in the 2024 presidential election, so maybe we can create a Black-Republican majority district there? Using data from Dave's Redistricting (thank you!) it is easy to show that there are 46 (!) precincts out of 2770 units. Impressive, Justice Alito!

000

Prof. Benjamin Forest @benjaminforest.bsky.social · 11/05/2026

In the 2024 election, there were nine counties that had both Black and Republican majorities, all very close to 50%. All were in the rural Deep South: Mississippi, Georgia, Alabama, and Louisiana. (Pop <40,000). Much more likely due to turnout differences than lots of Black voters supporting Trump.

041

Prof. Benjamin Forest @benjaminforest.bsky.social · 11/05/2026

How likely is it that a district with a Republican majority would also have a majority of Black voters? I compared the presidential results from 2016 and 2020 against the 2020 share of Black population by county. For both elections, there was exactly ONE county like this: Early County, Georgia.

A scatter plot showing the relationship between Republican support and Black population share by counties in the US for the 2016 and 2020 elections.

010

Prof. Benjamin Forest @benjaminforest.bsky.social · 17/06/2025

I got distracted thinking about turnout in US presidential elections. Here is a graph showing the total presidential votes from 1824-2024 as a percentage of eligible voters and as a percentage of total population.

020

Prof. Benjamin Forest @benjaminforest.bsky.social · 02/05/2025

To be clear, Canadians writing about Canada think this is always true.

020

Prof. Benjamin Forest @benjaminforest.bsky.social · 30/04/2025

The final results are in, per CBC, for vote share (and seats). LPC 43.7 (169). CPC 41.3 (144). Bloc: 6.3 (22). NDP 6.3 (7). Green 1.3 (1). Using these vote shares, the seats are high end of the forecast ranges for the CPC (126-148), Bloc (12-25) and Green (0-2).

220

Prof. Benjamin Forest @benjaminforest.bsky.social · 29/04/2025

The votes are still being counted but it appears the LPC is headed for a minority government with 168 seats. Based on the preliminary popular vote, my model over-predicted their seat count, but the CPC (144) is within the predicted range (albeit on the very high end).

110

Prof. Benjamin Forest @benjaminforest.bsky.social · 28/04/2025

RE: Trump's statement on the Canadian election, endorsing Pierre Poilievre without actually naming him, saying PP will help Canada become a state.

210

Prof. Benjamin Forest @benjaminforest.bsky.social · 28/04/2025

My forecasts based on the final CBC polling averages. The LPC are on track for a majority government with 190-191 seats with the CPC in opposition (132-133 seats). NDP (1-2), Bloc (17-18), and Greens (0-1) are far behind, but the NDP will probably do better at the expense of the LPC.

020

Prof. Benjamin Forest @benjaminforest.bsky.social · 26/04/2025

Polls still don’t show any major national swings. The CBC average has the LPC (42.5%) leading the CPC (38.7%) by about 4 percentage points, and 8 of the 10 most recent polls has them leading by at least 3 points. Today’s average forecasts 190-191 seats for the LPC and 130-131 for the CPC. 1/2

000

Prof. Benjamin Forest @benjaminforest.bsky.social · 22/04/2025

Although the absolute support for the LPC is lower in these two polls, it gives them the highest average seat count: 194.5, and a substantially worse result for the Conservatives (124 seats). These polls are also marginally better for the NDP (3.5 seats) and worse for the Bloc (18 seats). 2/2

010

Prof. Benjamin Forest @benjaminforest.bsky.social · 22/04/2025

The 40-40 tie from the Mainstreet poll still forecasts a Liberal victory, but with an average of only 181 seats and the CPC in opposition with 140-141 seats. Some simulations show a Liberal minority, but less than 1% of the time. As in 2021, the FPTP system helps the Liberals substantially. 4/4

000

Prof. Benjamin Forest @benjaminforest.bsky.social · 22/04/2025

The Angus Reid results have the Liberals averaging 190 seats and the Conservatives 129, with no simulation producing a minority government. The forecast has the Bloc with a solid 21-22 seats share and the now-familiar dire results for the NDP ( 1 seat or less).3/4

000

Prof. Benjamin Forest @benjaminforest.bsky.social · 19/04/2025

The model continues to forecast a comfortable majority for the Liberals, averaging 198 seats, with a 20% chance of cracking the 200-seat ceiling, and no chance of falling below a majority. 2/3

000

Prof. Benjamin Forest @benjaminforest.bsky.social · 10/04/2025

I ran the model using the 2021 boundaries (338 constituencies) with today’s CBC polling averages (10 April) to see what the likely outcome would be under the old system. The LPC averages 207 seats while the CPC averages 111. 3/6

010

Prof. Benjamin Forest @benjaminforest.bsky.social · 08/04/2025

There has been very little movement in the polling averages since 2 April, which is good news for the Liberals because they continue to hold a commanding lead in seats. The average count across the simulations is 201 seats, with CPC at 125.

010

Prof. Benjamin Forest @benjaminforest.bsky.social · 05/04/2025

In contrast, the model shows a more typical majority for the LPC when using the Mainstreet numbers, with an average of 187 seats, a high of less than 200, and one simulation (out of 10,000) even showing them one seat below a majority. 4/4

020

Prof. Benjamin Forest @benjaminforest.bsky.social · 05/04/2025

Anyway. . . The Nanos poll show the LPC with a commanding seat lead. The average is 210, and there is a small chance (<1%) of more than 220 seats. Similarly, there is <1% chance of fewer than 200 seats, which would be an unprecedented majority for them.

010

Prof. Benjamin Forest @benjaminforest.bsky.social · 02/04/2025

The LPC is up about 1.5% and the NDP down about 0.7% in CBC polling average compared to 31 March, but this is enough to move the forecast. The Liberals are now averaging 198 seats and the model shows a 15% chance of zero NDP seats.

030

Prof. Benjamin Forest @benjaminforest.bsky.social · 01/04/2025

A stunning new poll from Poisson d'Avril shows the election now as a two-party race. The Greens look to dominate the next government.

131

Prof. Benjamin Forest @benjaminforest.bsky.social · 31/03/2025

There are not a lot of changes from the 29 March results. The Liberals and NDP are up slightly, and the Bloc is down a bit in polls. The Liberals are maintaining their (forecasted) majority (average of 194 seats) and things are *slightly* less dire for the NDP (but still only averaging 2.5 seats).

020

Prof. Benjamin Forest @benjaminforest.bsky.social · 29/03/2025

This graph provides useful background for election projections. It shows the relationship between the national vote and seats won from 1945 to 2021. The bias usually works in favour of parties getting more than about 32% of the national vote. The 2021 bias favouring the Liberals was unusually large.

031

Prof. Benjamin Forest @benjaminforest.bsky.social · 29/03/2025

My forecast based on the CBC polling average for 29 March. The Liberals have extended their lead in the polls, which translates into a projected majority government in all simulations. The model has the LPC averaging 189 seats. Things look worse for the NDP, with zero seats in 5% of the simulations.

111

Prof. Benjamin Forest @benjaminforest.bsky.social · 26/03/2025

My updated forecast using the 26 March CBC polling averages has the Liberal Party winning a majority in almost every simulation, and an average of 184 seats across the 10,000 simulations. There are some dire outcomes for the NDP, with some simulations resulting in zero seats.

2228

Prof. Benjamin Forest @benjaminforest.bsky.social · 23/03/2025

The current CBC polling average has both the Liberals and Conservatives at about 37%, higher than both of their 2021 results (32.6 and 33.7 respectively). Using those figures, my model predicts a Liberal win, with a majority government as the most likely outcome.

2244

Prof. Benjamin Forest @benjaminforest.bsky.social · 10/12/2023

J'ai repris mon modèle swing pour le plus récent sondage québécois. La CAQ est en chute libre et le PQ en est le bénéficiaire. Selon ce sondage Léger, ils remporteraient en moyenne 61 à 62 sièges, soit près des 63 nécessaires pour obtenir une majorité.

020

Prof. Benjamin Forest @benjaminforest.bsky.social · 11/11/2023

I finally had a chance to run the population numbers on the newest Canadian and US federal election districts. The contrast in population equality standards is clear!

020